"Simpson Paradox" and Federer

[coban]

http://www.theatlantic.com/entertainment/archive/2014/01/what-every-pro-tennis-player-does-better-than-roger-federer/283007/

- ref above article, what do you guys think about this?

 

[GameSetAndMath]

http://www.theatlantic.com/entertainment/archive/2014/01/what-every-pro-tennis-player-does-better-than-roger-federer/283007/

- ref above article, what do you guys think about this?

Nice article. Thanks for posting it Coban.

Basically, this shows that Fed rarely wins ugly.

p.s. You might be interested in reading the "Random Walks and Tennis Scoring" thread
where there is an article focused on tennis scoring system.

 

[DarthFed]

http://www.theatlantic.com/entertainment/archive/2014/01/what-every-pro-tennis-player-does-better-than-roger-federer/283007/

- ref above article, what do you guys think about this?

Nice article. Thanks for posting it Coban.

Basically, this shows that Fed rarely wins ugly.

p.s. You might be interested in reading the "Random Walks and Tennis Scoring" thread
where there is an article focused on tennis scoring system.

All this really shows is that he isn't good in the tight matches, it is similar to the 5 set record. There is no way to put a positive spin on it, IMO. If you win more points than your opponent and lose the match this generally means you stunk in crunch time.

 

[Denis]

http://www.theatlantic.com/entertainment/archive/2014/01/what-every-pro-tennis-player-does-better-than-roger-federer/283007/

- ref above article, what do you guys think about this?

Nice article. Thanks for posting it Coban.

Basically, this shows that Fed rarely wins ugly.

p.s. You might be interested in reading the "Random Walks and Tennis Scoring" thread
where there is an article focused on tennis scoring system.

All this really shows is that he isn't good in the tight matches, it is similar to the 5 set record. There is no way to put a positive spin on it, IMO. If you win more points than your opponent and lose the match this generally means you stunk in crunch time.

Yeah, I was about to say the same. I bet there are a couple of matches against Nadal among them.

 

[GameSetAndMath]

http://www.theatlantic.com/entertainment/archive/2014/01/what-every-pro-tennis-player-does-better-than-roger-federer/283007/

- ref above article, what do you guys think about this?

Nice article. Thanks for posting it Coban.

Basically, this shows that Fed rarely wins ugly.

p.s. You might be interested in reading the "Random Walks and Tennis Scoring" thread
where there is an article focused on tennis scoring system.

All this really shows is that he isn't good in the tight matches, it is similar to the 5 set record. There is no way to put a positive spin on it, IMO. If you win more points than your opponent and lose the match this generally means you stunk in crunch time.

That is true. I agree. You lose a match because you suck in crunch time.

When you earn 20 break point chances and get 1 converted and the other gets
2 break point chances and converts both, then the other can win while you have
lot of points. This is what happens to Federer in many matches recently and
in many matches against Nadal in FO.

My comment of "This shows Fed wins ugly rarely" is orthogonal to the
material in this article and so is not relevant. That got to do with how much
percentage of matches are won by a player are won in an ugly manner.
The statistic in this article considers only those matches which are
examples of Simpson's paradox.

On the other hand there are also lot of people who think the mark
of champion is to win when not playing well. There is merit to that line
of thinking as well.

 

[Riotbeard]

Interesting stats. I don't buy the article's contention that this is further support that Fed is the best ever, even though I might concede that point in a general sense.

 

[DarthFed]

No, all it really shows is that his career could have been a whole hell of a lot greater when all is said and done. Of course there is no breakdown as to how many of those matches came when he was younger (before he was a contender at slams) and how many of those matches are at slams to begin with. I know that Fed won more points than Nadal at 2009 AO final. Wouldn't be surprised if that was the case vs. Safin at 05 AO as well. Not sure of others offhand.

 

[Moxie]

No, all it really shows is that his career could have been a whole hell of a lot greater when all is said and done. Of course there is no breakdown as to how many of those matches came when he was younger (before he was a contender at slams) and how many of those matches are at slams to begin with. I know that Fed won more points than Nadal at 2009 AO final. Wouldn't be surprised if that was the case vs. Safin at 05 AO as well. Not sure of others offhand.

So you and GSM aren't buying the author's notion that the stat might come from a lot of matches where players had to go for broke against Roger? It's not like Fed has won so many in nail-biters, or in matches where some might say "he shouldn't have won it." I do buy the idea that it means he was confident enough to bide his time, and play the important points well. Like Riotbeard, however, I would say I don't see why this argument buffs up his claim to GOAT, though there are other (and better) reasons to support it.

 

[Tennis Miller]

I'm not sure this small sample size tells us much either way about Federer, who's played, what, like 1100 matches. What about matches where he won the same number of points, or just a few more, than the other guy. Wouldn't those be included in assessing how well he does in close matches? I don't think this "paradox" would tell us much about any top player actually, with a similarly small sample size.

 

[GameSetAndMath]

http://www.theatlantic.com/entertainment/archive/2014/01/what-every-pro-tennis-player-does-better-than-roger-federer/283007/

- ref above article, what do you guys think about this?

Nice article. Thanks for posting it Coban.

Basically, this shows that Fed rarely wins ugly.

p.s. You might be interested in reading the "Random Walks and Tennis Scoring" thread
where there is an article focused on tennis scoring system.

All this really shows is that he isn't good in the tight matches, it is similar to the 5 set record. There is no way to put a positive spin on it, IMO. If you win more points than your opponent and lose the match this generally means you stunk in crunch time.

Actually, we cannot even conclude from this that he isn't good in tight matches. As this
does not take into account all the tight matches. The problem is that tight matches is not
the same as matches involving Simpson's paradox. For example, a 7-6, 6-7, 7-6 match
won by Fed can certainly be called a tight match. But, if in that match Fed has won 120
points and his opponent has won 119 points, then it does not create Simpson's
paradox and so is not included in the list of matches studied by the article. This is
what Tennis Miller also alludes in his post.

At this time, I am not really ready to say anything about the meaning of the
statistic generated by the authors of the article. I need to think more about it.

 

[GameSetAndMath]

While this is an interesting article, I think it is best to leave the GOAT out of this
discussion (even though article mentions it in the conclusion) as otherwise this
may degenerate into familiar path GOAT discussion takes.

 

[GameSetAndMath]

I'm not sure this small sample size tells us much either way about Federer, who's played, what, like 1100 matches. What about matches where he won the same number of points, or just a few more, than the other guy. Wouldn't those be included in assessing how well he does in close matches? I don't think this "paradox" would tell us much about any top player actually, with a similarly small sample size.

Although not required by the scoring mechanism, most of the time the player with
more points wins the match also. Let us call a win a lucky win if it is an example of
Simpson's paradox (although it need not necessarily be lucky).

One way to interpret the results of the article is that if a match involving Fed
is won by someone in a lucky manner, it is more often won by Fed's opponent than Fed.
In other words, Fed is more often unlucky than lucky.

It would be interesting if the author can release the complete list of such
matches involving Fed. I wonder whether 5-set loss to Djokovic (where
Djoker took a wild swing in return on an important point) at USO falls in
this category.

 

[Moxie]

http://www.theatlantic.com/entertainment/archive/2014/01/what-every-pro-tennis-player-does-better-than-roger-federer/283007/

- ref above article, what do you guys think about this?

Nice article. Thanks for posting it Coban.

Basically, this shows that Fed rarely wins ugly.

p.s. You might be interested in reading the "Random Walks and Tennis Scoring" thread
where there is an article focused on tennis scoring system.

All this really shows is that he isn't good in the tight matches, it is similar to the 5 set record. There is no way to put a positive spin on it, IMO. If you win more points than your opponent and lose the match this generally means you stunk in crunch time.

Actually, we cannot even conclude from this that he isn't good in tight matches. As this
does not take into account all the tight matches. The problem is that tight matches is not
the same as matches involving Simpson's paradox. For example, a 7-6, 6-7, 7-6 match
won by Fed can certainly be called a tight match. But, if in that match Fed has won 120
points and his opponent has won 119 points, then it does not create Simpson's
paradox and so is not included in the list of matches studied by the article. This is
what Tennis Miller also alludes in his post.

At this time, I am not really ready to say anything about the meaning of the
statistic generated by the authors of the article. I need to think more about it.

I agree with you, GSM, that it wants a little more thinking. What I think Tennis Miller, and DarthFed and some other Fed fans are taking too much on the chin is the notion that the statistic shows Roger up poorly in tight matches. I don't actually think that's what those numbers mean. And clearly that was not the author's point. But he has played some 1000-1100 matches in his career, or thereabouts, so there can be all kinds of crazy statistics within them. If the basic notion is that he has won more matches than anyone who has not won the most points, it seems, on face, that he's MORE clutch, not less.

 

[DarthFed]

I'm not sure this small sample size tells us much either way about Federer, who's played, what, like 1100 matches. What about matches where he won the same number of points, or just a few more, than the other guy. Wouldn't those be included in assessing how well he does in close matches? I don't think this "paradox" would tell us much about any top player actually, with a similarly small sample size.

Although not required by the scoring mechanism, most of the time the player with
more points wins the match also. Let us call a win a lucky win if it is an example of
Simpson's paradox (although it need not necessarily be lucky).

One way to interpret the results of the article is that if a match involving Fed
is won by someone in a lucky manner, it is more often won by Fed's opponent than Fed.
In other words, Fed is more often unlucky than lucky.

It would be interesting if the author can release the complete list of such
matches involving Fed. I wonder whether 5-set loss to Djokovic (where
Djoker took a wild swing in return on an important point) at USO falls in
this category.

Djokovic had more points than Roger in the 2011 semi, not sure about 2010 though I'd think Djokovic would have more points since he won 2 lopsided sets in that one.

You and Tennis Miller are right that it does not include matches where the winner won a few more points than the loser (obviously a tight match). So it is incomplete, yet 4-24 is rather overwhelming. It just means in those matches on average Roger did not do well in big moments. Taking a match like 2009 AO that was obviously the case. I wonder if 2009 USO is also one of those 24 losses. Not sure what else to read into that stat. In no way does it make him look good unless I'm completely blind.

 

[DarthFed]

And not to switch gears but a much more important and relevant statistic is the 5 set record, especially near the end against quality competition. Roger is 3-7 in 5 set matches in the semis and finals of majors. That's why I mentioned that this particular stat is kind of on cue with that one. No one is perfect obviously...if Roger was good on that stat he would damn near be perfect though.

 

[Murat Baslamisli]

And not to switch gears but a much more important and relevant statistic is the 5 set record, especially near the end against quality competition. Roger is 3-7 in 5 set matches in the semis and finals of majors. That's why I mentioned that this particular stat is kind of on cue with that one. No one is perfect obviously...if Roger was good on that stat he would damn near be perfect though.

Roger would probably say " If any of my matches go to 5 sets, that means I was not playing well that day anyways"

 

[DarthFed]

And not to switch gears but a much more important and relevant statistic is the 5 set record, especially near the end against quality competition. Roger is 3-7 in 5 set matches in the semis and finals of majors. That's why I mentioned that this particular stat is kind of on cue with that one. No one is perfect obviously...if Roger was good on that stat he would damn near be perfect though.

Roger would probably say " If any of my matches go to 5 sets, that means I was not playing well that day anyways"

That's what I'd say too but that talk often got me in a lot of "heated discussions" with MikeOne and one of his 100,000 alter egos.

 

[Moxie]

I'm not sure this small sample size tells us much either way about Federer, who's played, what, like 1100 matches. What about matches where he won the same number of points, or just a few more, than the other guy. Wouldn't those be included in assessing how well he does in close matches? I don't think this "paradox" would tell us much about any top player actually, with a similarly small sample size.

Although not required by the scoring mechanism, most of the time the player with
more points wins the match also. Let us call a win a lucky win if it is an example of
Simpson's paradox (although it need not necessarily be lucky).

One way to interpret the results of the article is that if a match involving Fed
is won by someone in a lucky manner, it is more often won by Fed's opponent than Fed.
In other words, Fed is more often unlucky than lucky.

It would be interesting if the author can release the complete list of such
matches involving Fed. I wonder whether 5-set loss to Djokovic (where
Djoker took a wild swing in return on an important point) at USO falls in
this category.

Djokovic had more points than Roger in the 2011 semi, not sure about 2010 though I'd think Djokovic would have more points since he won 2 lopsided sets in that one.

You and Tennis Miller are right that it does not include matches where the winner won a few more points than the loser (obviously a tight match). So it is incomplete, yet 4-24 is rather overwhelming. It just means in those matches on average Roger did not do well in big moments. Taking a match like 2009 AO that was obviously the case. I wonder if 2009 USO is also one of those 24 losses. Not sure what else to read into that stat. In no way does it make him look good unless I'm completely blind.

I think you ARE completely blind, in your pessimism. Don't compare him to the likes of Djokovic or Nadal in such a stat. That's clearly not where he picked up the most of them. It's against under-card players going for broke. The stat's not supposed to make him look bad, it's supposed to make him look good. I completely disagree that it says that Roger didn't do well in the big moments. Exactly the opposite, I'd say: it means he DID play the big moments well. That's why he could let garbage points go by. Because he'd still get over on the opponent, when it mattered.

 

[DarthFed]

I'm not sure this small sample size tells us much either way about Federer, who's played, what, like 1100 matches. What about matches where he won the same number of points, or just a few more, than the other guy. Wouldn't those be included in assessing how well he does in close matches? I don't think this "paradox" would tell us much about any top player actually, with a similarly small sample size.

Although not required by the scoring mechanism, most of the time the player with
more points wins the match also. Let us call a win a lucky win if it is an example of
Simpson's paradox (although it need not necessarily be lucky).

One way to interpret the results of the article is that if a match involving Fed
is won by someone in a lucky manner, it is more often won by Fed's opponent than Fed.
In other words, Fed is more often unlucky than lucky.

It would be interesting if the author can release the complete list of such
matches involving Fed. I wonder whether 5-set loss to Djokovic (where
Djoker took a wild swing in return on an important point) at USO falls in
this category.

Djokovic had more points than Roger in the 2011 semi, not sure about 2010 though I'd think Djokovic would have more points since he won 2 lopsided sets in that one.

You and Tennis Miller are right that it does not include matches where the winner won a few more points than the loser (obviously a tight match). So it is incomplete, yet 4-24 is rather overwhelming. It just means in those matches on average Roger did not do well in big moments. Taking a match like 2009 AO that was obviously the case. I wonder if 2009 USO is also one of those 24 losses. Not sure what else to read into that stat. In no way does it make him look good unless I'm completely blind.

I think you ARE completely blind, in your pessimism. Don't compare him to the likes of Djokovic or Nadal in such a stat. That's clearly not where he picked up the most of them. It's against under-card players going for broke. The stat's not supposed to make him look bad, it's supposed to make him look good. I completely disagree that it says that Roger didn't do well in the big moments. Exactly the opposite, I'd say: it means he DID play the big moments well. That's why he could let garbage points go by. Because he'd still get over on the opponent, when it mattered.

First of all we don't know what matches make up this stat. The author of the article is taking a guess as to the reason the record is so poor in that situation but for all we know a lot of these matches were vs. top 10. In fact, that is most likely the case because he was rarely in close contests with lower ranked players until very recently.

Losing 24 times where he won more points than his opponent vs. winning 4 times where he won less points somehow shows that he played well in the tight moments in those 28 matches? I'm confused :nono

 

[Moxie]

I'm not sure this small sample size tells us much either way about Federer, who's played, what, like 1100 matches. What about matches where he won the same number of points, or just a few more, than the other guy. Wouldn't those be included in assessing how well he does in close matches? I don't think this "paradox" would tell us much about any top player actually, with a similarly small sample size.

Although not required by the scoring mechanism, most of the time the player with
more points wins the match also. Let us call a win a lucky win if it is an example of
Simpson's paradox (although it need not necessarily be lucky).

One way to interpret the results of the article is that if a match involving Fed
is won by someone in a lucky manner, it is more often won by Fed's opponent than Fed.
In other words, Fed is more often unlucky than lucky.

It would be interesting if the author can release the complete list of such
matches involving Fed. I wonder whether 5-set loss to Djokovic (where
Djoker took a wild swing in return on an important point) at USO falls in
this category.

Djokovic had more points than Roger in the 2011 semi, not sure about 2010 though I'd think Djokovic would have more points since he won 2 lopsided sets in that one.

You and Tennis Miller are right that it does not include matches where the winner won a few more points than the loser (obviously a tight match). So it is incomplete, yet 4-24 is rather overwhelming. It just means in those matches on average Roger did not do well in big moments. Taking a match like 2009 AO that was obviously the case. I wonder if 2009 USO is also one of those 24 losses. Not sure what else to read into that stat. In no way does it make him look good unless I'm completely blind.

I think you ARE completely blind, in your pessimism. Don't compare him to the likes of Djokovic or Nadal in such a stat. That's clearly not where he picked up the most of them. It's against under-card players going for broke. The stat's not supposed to make him look bad, it's supposed to make him look good. I completely disagree that it says that Roger didn't do well in the big moments. Exactly the opposite, I'd say: it means he DID play the big moments well. That's why he could let garbage points go by. Because he'd still get over on the opponent, when it mattered.

First of all we don't know what matches make up this stat. The author of the article is taking a guess as to the reason the record is so poor in that situation but for all we know a lot of these matches were vs. top 10. In fact, that is most likely the case because he was rarely in close contests with lower ranked players until very recently.

Losing 24 times where he won more points than his opponent vs. winning 4 times where he won less points somehow shows that he played well in the tight moments in those 28 matches? I'm confused :nono

To be honest, perhaps I'm confused, too. I thought the article stated the opposite. Too many double-negatives, perhaps. But if I'm wrong, why is the author trying to make this as an argument for Roger's greatness? That wouldn't make sense.